encode.py
8.53 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
'''
Created on Oct 23, 2013
@author: mlenart
'''
from morfeuszbuilder.utils.serializationUtils import htons, serializeString
from morfeuszbuilder.utils import limits, exceptions
MAX_QUALIFIERS_COMBINATIONS = 1024
class Encoder(object):
'''
classdocs
'''
def __init__(self, lowercase, encoding='utf8'):
'''
Constructor
'''
self.lowercase = lowercase
self.encoding = encoding
#~ self.qualifiersMap = { frozenset(): 0}
def encodeWord(self, word, lowercase=True):
assert type(word) == str
res = bytearray(word.lower() if self.lowercase and lowercase else word, self.encoding)
return res
def encodeData(self, data):
raise NotImplementedError()
def decodeData(self, rawData):
return NotImplementedError()
def decodeWord(self, rawWord):
return str(str(rawWord).strip('\x00'), self.encoding)
def word2SortKey(self, word):
normalizedWord = word.lower() if self.lowercase else word
return normalizedWord
def _encodeTypeNum(self, typenum):
exceptions.validate(
typenum <= limits.MAX_SEGMENT_TYPES,
'Too many segment types. The limit is %d' % limits.MAX_SEGMENT_TYPES)
return bytearray([typenum])
def _hasUpperPrefix(self, casePattern):
for i in range(len(casePattern) + 1):
if all(casePattern[:i]) and not any(casePattern[i:]):
return True
return False
def _getUpperPrefixLength(self, casePattern):
assert self._hasUpperPrefix(casePattern)
for i in range(len(casePattern)):
if not casePattern[i]:
return i
return len(casePattern)
def _encodeTagNum(self, tagnum):
res = bytearray()
exceptions.validate(tagnum <= limits.MAX_TAGS, 'Too many tags. The limit is %d' % limits.MAX_TAGS)
res.append((tagnum & 0xFF00) >> 8)
res.append(tagnum & 0x00FF)
return res
def _encodeNameNum(self, namenum):
exceptions.validate(namenum <= limits.MAX_NAMES, 'Too many named entity types. The limit is %d' % limits.MAX_NAMES)
return bytearray([namenum])
def _groupInterpsByType(self, interpsList):
res = {}
for interp in interpsList:
res.setdefault(interp.typenum, [])
res[interp.typenum].append(interp)
return res
def _doEncodeData(self, interpsList):
assert type(interpsList) == frozenset
segnum2Interps = self._groupInterpsByType(interpsList)
res = bytearray()
for typenum, interpsList in list(segnum2Interps.items()):
res.extend(self._encodeInterps4Type(typenum, interpsList))
del interpsList
res = htons(len(res)) + res
return res
class MorphEncoder(Encoder):
def __init__(self, encoding='utf8'):
super(MorphEncoder, self).__init__(True, encoding)
def encodeData(self, interpsList):
return self._doEncodeData(interpsList)
def _getMinOrthCasePatterns(self, interpsList):
res = []
for interp in interpsList:
if not True in interp.orthCasePattern:
return []
else:
res.append(list(interp.orthCasePattern))
return res
def _encodeCasePattern(self, casePattern):
LEMMA_ONLY_LOWER = 0
LEMMA_UPPER_PREFIX = 1
LEMMA_MIXED_CASE = 2
res = bytearray()
if True not in casePattern:
res.append(LEMMA_ONLY_LOWER)
return res
elif self._hasUpperPrefix(casePattern):
res.append(LEMMA_UPPER_PREFIX)
res.append(self._getUpperPrefixLength(casePattern))
return res
else:
assert len(casePattern) < 256
res.append(LEMMA_MIXED_CASE)
res.append(len([c for c in casePattern if c]))
for idx in range(len(casePattern)):
if casePattern[idx]:
res.append(idx)
return res
def _casePatternsHaveOnlyLowercase(self, casePatterns):
return not any([cp and True in cp for cp in casePatterns])
def _casePatternsAreOnlyTitles(self, casePatterns):
return all([cp and cp[0] == True and not True in cp[1:] for cp in casePatterns])
def _casePatternsAreEncodedInCompressByte(self, casePatterns):
return self._casePatternsHaveOnlyLowercase(casePatterns) or self._casePatternsAreOnlyTitles(casePatterns)
def _prefixCutsAreEncodedInCompressByte(self, prefixCuts):
return len(prefixCuts) == 1 and list(prefixCuts)[0] < 15
def _encodeCompressByte(self, orthCasePatterns, lemmaCasePatterns, prefixCuts):
ORTH_ONLY_LOWER = 128
ORTH_ONLY_TITLE = 64
LEMMA_ONLY_LOWER = 32
LEMMA_ONLY_TITLE = 16
PREFIX_CUT_MASK = 15
res = 0
if self._casePatternsHaveOnlyLowercase(orthCasePatterns):
res |= ORTH_ONLY_LOWER
elif self._casePatternsAreOnlyTitles(orthCasePatterns):
res |= ORTH_ONLY_TITLE
if self._casePatternsHaveOnlyLowercase(lemmaCasePatterns):
res |= LEMMA_ONLY_LOWER
elif self._casePatternsAreOnlyTitles(lemmaCasePatterns):
res |= LEMMA_ONLY_TITLE
if self._prefixCutsAreEncodedInCompressByte(prefixCuts):
res |= list(prefixCuts)[0]
else:
res |= PREFIX_CUT_MASK
return res
def _encodeInterps4Type(self, typenum, interpsList):
res = bytearray()
res.extend(self._encodeTypeNum(typenum))
encodedInterpsList = bytearray()
orthCasePatterns = set([tuple(interp.orthCasePattern) for interp in interpsList])
lemmaCasePatterns = set([tuple(interp.encodedForm.casePattern) for interp in interpsList])
prefixCuts = set([interp.encodedForm.prefixCutLength for interp in interpsList])
encodedInterpsList.append(self._encodeCompressByte(orthCasePatterns, lemmaCasePatterns, prefixCuts))
if not self._casePatternsAreEncodedInCompressByte(orthCasePatterns):
minOrthCasePatterns = self._getMinOrthCasePatterns(interpsList)
encodedInterpsList.append(len(minOrthCasePatterns))
for casePattern in minOrthCasePatterns:
encodedInterpsList.extend(self._encodeCasePattern(casePattern))
for interp in sorted(interpsList, key=lambda i: i.getSortKey()):
if not self._casePatternsAreEncodedInCompressByte(orthCasePatterns):
encodedInterpsList.extend(self._encodeCasePattern(interp.orthCasePattern))
if not self._prefixCutsAreEncodedInCompressByte(prefixCuts):
encodedInterpsList.append(interp.encodedForm.prefixCutLength)
encodedInterpsList.append(interp.encodedForm.cutLength)
encodedInterpsList.extend(serializeString(interp.encodedForm.suffixToAdd))
if not self._casePatternsAreEncodedInCompressByte(lemmaCasePatterns):
encodedInterpsList.extend(self._encodeCasePattern(interp.encodedForm.casePattern))
encodedInterpsList.extend(htons(interp.tagnum))
encodedInterpsList.append(interp.namenum)
encodedInterpsList.extend(htons(interp.qualifiers))
res.extend(htons(len(encodedInterpsList)))
res.extend(encodedInterpsList)
return res
class Encoder4Generator(Encoder):
def __init__(self, encoding='utf8'):
super(Encoder4Generator, self).__init__(False, encoding)
def encodeData(self, interpsList):
return self._doEncodeData(interpsList)
def _encodeInterps4Type(self, typenum, interpsList):
res = bytearray()
res.extend(self._encodeTypeNum(typenum))
encodedInterpsList = bytearray()
for interp in sorted(interpsList, key=lambda i: i.getSortKey()):
encodedInterpsList.extend(serializeString(interp.homonymId))
encodedInterpsList.extend(serializeString(interp.encodedForm.prefixToAdd))
encodedInterpsList.append(interp.encodedForm.cutLength)
encodedInterpsList.extend(serializeString(interp.encodedForm.suffixToAdd))
encodedInterpsList.extend(htons(interp.tagnum))
encodedInterpsList.append(interp.namenum)
encodedInterpsList.extend(htons(interp.qualifiers))
res.extend(htons(len(encodedInterpsList)))
res.extend(encodedInterpsList)
return res